Investigation on Time Domain Signal to Appropriate Wavelet Transform Threshold Function Selection for Partial Discharge Detection

Dielectric insulation is mostly affected due to Partial discharge (PD) pulses, which are resultants of various stress parameters. Normal acquired signal from any insulating material has two components namely actual partial discharge (PD) pulses and noise. Noise suppression is crucial before any (PD) data analysis. PD pulses are short transient, randomly occurred and non-stationary signals. Recent research shows that the Wavelet transform (WT) has achieved good effect in noise rejection with acquired signal. WT proves its suitability by providing information in time and frequency domains simultaneously. Hence, WT can detect these pulses efficiently. PD signals extraction using WT requires set up of various parameters like, optimal mother wavelet selection, decomposition level selection, threshold rule and function determination. This paper presents the vital issues of WT threshold function determination. Here, a method is envisaged for appropriate WT threshold function selection based on acquired signal analysis. This work is carried out on signal measured from transformer. Also, analysis is carried out on the acquired signal and its WT generated signal. Finally, a comparison is carried out to select optimum threshold function selection for WT signal analysis.